Cohort identification

The cohort consisted of KPNC members who completed a survey as part of the Kaiser Permanente Research Program on Genes, Environment, and Health (RPGEH) between 2007 and 2009 and subsequently delivered a live-born singleton at KPNC between 2007 and 2010. Pregnancies were identified through the KPNC Pregnancy Glucose Tolerance Registry, an ongoing registry that identifies all pregnancies reaching the second trimester and that has previously been described in detail.

The RPGEH survey was initially mailed to adult members (aged >18 years) of KPNC in February 2007. Approximately 400,000 members completed the survey and provided consent for linking the survey data to their electronic medical record (EMR), of which 57,100 were reproductive-aged women. The survey contained detailed information on pregravid factors including body mass index (BMI), alcohol consumption, medical and reproductive history, smoking, diet, and physical activity.

In general, this cohort of RPGEH survey responders who became pregnant was representative of the pregnancies occurring within Kaiser Permanente Northern California membership: 53% of the RPGEH cohort were from racial/ethnic minority groups compared with 58% among all KPNC pregnancies. There was slightly lower representation of African Americans (4.4% vs 6.6% in the KPNC pregnant population).

Among women who completed the RPGEH survey and had a subsequent pregnancy in KPNC, we identified 5967 mother/child pairs resulting in a singleton live birth with a child’s height and weight measurements taken at 13 months of age or older. If a woman had more than 1 pregnancy, we selected the first one after the RPGEH survey. We then restricted to mother/child pairs in which the child was aged 2 years or older at the time of the most recent weight and height measurement (n = 4505). We excluded pairs if the mother was missing data on gestational weight gain (n = 4151). We then limited the cohort to those who had complete data on gestational age at delivery, resulting in a final analytic cohort of 4145 mother/child pairs.

This study was approved by the Kaiser Foundation Research Institute Institutional Review Board and the State of California Committee for the Protection of Human Subjects.

Maternal characteristics

We obtained the following maternal characteristics from the RPGEH survey: self-reported maternal race/ethnicity ([1] non-Hispanic white, [2] African American, [3] Asian, [4] Hispanic, and [5] unknown) and educational attainment (in years).

Prepregnancy dietary pattern

The RPGEH survey included 20 food categories, including sugar-sweetened beverages. To identify major prepregnancy dietary patterns, principal components analysis was used on the 20 foods to identify factors that accounted for much of the variance. The food groups (factors) were rotated using an orthogonal transformation, resulting in uncorrelated, independent factors. The factor score for each factor (pattern) was calculated by summing intakes of food groups weighted by factor loading, and each individual was assigned a score for each identified pattern. Individuals with a high score for a pattern compared with individuals with lower scores have a stronger tendency to follow that pattern. The scores were then categorized by tertiles.

We identified 2 distinct dietary patterns: prudent and Western. Prudent diet is characterized by consumption of the following food groups: vegetables, fruits, whole grains, nuts/seeds, and beans. A Western diet is characterized by the consumption of processed meats, baked goods, whole eggs, beef, pork, lamb, deep-fried foods, margarine, and soft drinks.

Prepregnancy physical activity

The volume of total metabolic equivalent (METs) was calculated as minutes per week based on 4 questions.

Gestational diabetes

Gestational diabetes (GDM) was assessed through the KPNC Pregnancy Glucose Tolerance Registry and defined as having at least 2 plasma glucose values on the 100 g, 3 hour oral glucose tolerance test meeting or exceeding the Carpenter-Coustan thresholds (fasting: 95 mg/dL, 1 hour: 180 mg/dL, 2 hours: 155 mg/dL, 3 hours: 140 mg/dL).

Exposure ascertainment

Prepregnancy weight was a measured weight obtained before pregnancy (within 12 months of the last menstrual period) when available in the EMR (90%) or self-reported prepregnancy weight assessed at the first prenatal visit, also assessed via EMR. Prepregnancy BMI was calculated as prepregnancy weight (kilograms) divided by height (meters) squared. BMI categories were created in accordance with the 2009 IOM gestational weight gain recommendations as follows: underweight (<18.5 kg/m ² ), normal weight (18.5-24.9 kg/m ² ), overweight (25.0-29.9 kg/m ² ), and obese (≥30.0 kg/m ² ). Given the small proportion of women who were underweight (1.9%), the underweight and normal-weight categories were combined into 1 category (<24.9 kg/m ² ).

Total gestational weight gain was calculated as the difference between the last measured pregnancy weight and prepregnancy weight, in kilograms. The last pregnancy weight, obtained from the EMR, was the final predelivery weight and had to be measured no more than 4 weeks before delivery to be included in the analysis. Total gestational weight gain was categorized according to the 2009 IOM gestational weight gain recommendations (below, met, exceeded). The total rate of gestational weight gain per week during pregnancy (in kilograms) was calculated as the total gestational weight gain divided by the weeks of gestation attained at the last pregnancy weight measurement.

Offspring characteristics

Birthweight for gestational age was categorized according to the study population’s race/ethnicity- and gestational age–specific birthweight distribution as follows: large for gestational age if birth weight greater than the 90th percentile, small for gestational age if birthweight was less than the 10th percentile, and appropriate for gestational age if birthweight between the 10th and 90th percentiles, inclusive. Sex, height, and weight data for the children were obtained from the EMR. Children were classified as overweight or obese if their BMI z-score met or exceeded the 85th percentile of the Centers for Disease Control and Prevention child growth standards, based on age and sex. International Classification of Diseases , ninth revision, delivery codes (764.x or 565.x [possible intrauterine growth restriction]) in our EMR were used to identify infants with intrauterine growth restriction.

Statistical analysis

Unconditional logistic regression analysis was used to obtain odds ratios (ORs) and confidence intervals (CIs) estimating the odds of subsequent childhood overweight-obesity associated with gestational weight gain. For models examining the IOM recommendations as the exposure, we estimated the odds of childhood overweight/obesity associated with exceeding and gaining below the recommendations, as compared with meeting the recommendations. For total gestational weight gain per week, we also categorized women into tertiles; those in the lowest tertile served as the reference group.

Variables evaluated for confounding included those of a priori interest (maternal age at delivery, race/ethnicity, prepregnancy BMI (kilograms per square meter), child age, preterm birth (<37 weeks’ gestation), maternal education, parity, infant sex, volume of physical activity [mean MET minutes per week], and maternal prudent pattern diet).

To assess confounding, we entered covariates into a logistic regression model one at a time and then compared the adjusted and unadjusted ORs. Final logistic regression models included the variables that were evaluated for confounding and further adjusted for birthweight and maternal pregnancy glycemia (defined as GDM [yes or no]), which may be mediators of the gestational weight gain and childhood overweight/obesity association.

We initially examined associations stratified by the 4 levels of BMI category used for the IOM recommendations; however, given that results were similar for overweight and obese women when stratifying by prepregnancy BMI, the 2 categories were combined. In addition, we had a very small percentage of women who were in the underweight category (1.9%).

To assess the potential modifying effects of prepregnancy BMI (≥25 kg/m ² vs <25 kg/m ² ), we included appropriate cross-product (interaction) terms in regression models. Finally, the following sensitivity analyses were conducted: (1) restricting to those with an appropriate for gestational age infant, and (2) excluding those with self-reported weights. SAS version 9.1 (SAS Institute Inc, Cary, NC) was used for all analyses.